|Publication number||US8018860 B1|
|Application number||US 10/387,289|
|Publication date||Sep 13, 2011|
|Filing date||Mar 12, 2003|
|Priority date||Mar 12, 2003|
|Publication number||10387289, 387289, US 8018860 B1, US 8018860B1, US-B1-8018860, US8018860 B1, US8018860B1|
|Inventors||Fred S. Cook|
|Original Assignee||Sprint Communications Company L.P.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (12), Non-Patent Citations (2), Referenced by (26), Classifications (7), Legal Events (4)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present invention relates in general to managing mixed-vendor, meshed communication networks, and, more specifically, to predicting re-route interactions in association with maintenance actions on network elements.
With the increasing complexity of communication networks, such as layered protocol networks (IP over ATM, or IP over Frame Relay) used with the Internet and synchronous optical networks (SONET) used in telephone networks, network management functions such as network element provisioning and configuration, load monitoring, and fault detection become increasingly important. Currently, network element management systems collect network element resource status via various management protocols including TL1, CMIP, CORBA, SNMP, and others. Network element resources may, for example, include routers, switches, servers, bridges, multiplexers, and other devices depending upon the type of network. The availability of element resource status allows external (i.e., management) systems to determine the load and utilization levels of the device, the fault state (or impending fault states), and current configurations. This collected information allows network service providers to perform the standard FCAPS (fault, configuration, accounting, performance, and security) functions associated with managing a network.
Recently, it has become more and more popular to relegate the control portion of the network element configuration function of FCAPS to a logically separate “control plane” system. In some cases, the control functionality is housed within the network element itself, and, in other cases, it is housed in near proximity to the network element. The control functionality is concerned principally with fulfilling explicit or implicit end-user requests (e.g., requests whose response time is clearly discernable by the end-user). These functions typically involve providing a transient connection or allocating processing or storage resources to a specific user request. Fault detection, correction, and restoration after failure of these resources are also typically handled by the control plane system.
Traffic load levels within a network impact the performance of all network elements. To maintain a reasonable system cost, networks are typically over-subscribed for their potential peak traffic rates. In other words, the available resources of the network could not support all possible requests that could potentially occur at the same time (e.g., the telephone network cannot support a situation where every telephone is in use simultaneously).
In a meshed network, each network element is connected to many other network elements so that network traffic can potentially reach its destination by many different paths. Due to the large size and complexity of most networks, network elements from a variety of vendors/manufacturers are typically present. Unfortunately, when pieces of the control functionality are shared across multiple vendors and/or element types, the restoration steps taken for a resource becomes unpredictable. Since the various vendors and/or network element types do not have an agreed upon standard method of restoration between themselves, restoration actions must be coordinated above the network element level to be rational and predictable. Restoration that is coordinated external to the network element level is frequently too slow to fall within the service level agreement (SLA) allowances.
When a failure of a network element or other error occurs making a communication path in a meshed network unavailable, the traffic that was being handled by a number of transport paths, x, must then be handled by x−1 paths. In an IP network, for example, the error correction action (i.e., re-convergence) automatically re-routes traffic paths over the remaining links after some amount of convergence time. This process, however, does not take SLA parameters into consideration when determining how paths are re-routed. Consequently, for a premium network service such as video conferencing, SLA requirements for a limited transport latency and/or jitter may be violated by the newly converged configuration and/or by the re-convergence action itself (e.g. re-convergence takes 10 minutes on an SLA budget that allows 5 minutes outage annually).
Some network elements attempt to lessen these problems by allowing an operator to provision a failover resource to take over as a backup when the main resource becomes unavailable. In that case, no communication is required between network elements when a failure occurs—network operation merely switches from the failed resource to the provisioned backup resource. Manual provisioning of failover resources is undesirable in a typical network, however, because thousands of network element resources are present and it burdensome to manually configure and/or reconfigure all these resources. In addition, providing a failover resource for each network element doubles the resource requirements (and cost) of the network, or configured failover resources are re-used which increases the likelihood that failover resources will already be in use when they are needed.
A typical method for reducing the number of resources required for failover protection is to provide one failover resource per each group of n resources, where n is the number of network element resources in a group to be served by one failover resource. This tremendously improves the resource utilization of the network, but has led to other problems. More specifically, when either 1) multiple network elements fail simultaneously, 2) operation of two or more network elements is suspended while performing maintenance actions on the network elements, or 3) a network element fails while another network element is down for maintenance, then multiple requests for the same failover resource can occur (e.g., two or more resources of the “n” group are out of service at the same time and traffic from both is switched to the same failover device). For instance, if two operators independently perform maintenance on two different resources in a network, it is possible that the re-routing generated by the out-of-service resources will failover to the same alternate resource at some point in the network. Furthermore, the number of nodes (i.e., elements), the number of connections between the nodes (i.e., links), and the number of virtual paths traversing the nodes prevents a network operator from understanding the likely interactions created as a result of any particular failover. As a result, network performance is degraded and may result in noncompliance with the provider's SLA, or worse yet, a cascade failure scenario.
There are currently network analysis tools (such as the Conscious™ traffic engineering system available from Zvolve Systems, Inc.) that will analyze element load levels and resource allocations based upon information retrieved from element management system data. These tools will then generate suggested resource allocation for any new requests, and they can be used to predict SLA violations for resource outages by adding requests for resources currently used by the failed element to a model of the network not including that element. Unfortunately, these tools do not predict what a network element will attempt to do when a failover occurs. This results in a situation where unpredicted and/or uncoordinated actions take place in a failure situation, and where the response to a failure situation is unacceptably slow (e.g., the network tool identifies an error and re-configures the network elements based upon its observations, in a process that takes from minutes to hours).
The present invention has the advantages that the effect of a link or device failure in a mixed-vendor, multi-node (typically a meshed) network is predicted and a network operator can know in advance whether a desired maintenance action is allowable given the current network loading, link, device status, and failover routing. Furthermore, failover settings for network elements can be better engineered to avoid undesirable re-routing interactions during maintenance or failure of network elements (i.e. failover routes are logically exercised to determine unacceptable re-use patterns and loading conditions).
In one aspect of the invention, a method is provided for predicting re-routing interactions in a communications network including a plurality of network elements. A respective device state image is constructed for each of the plurality of network elements. The device state images are transmitted to a network simulator. The performance of the communications network is simulated in response to the device state images. A prospective network element failure of a predetermined one of the network elements is transmitted to the network simulator. A performance of the communications network is simulated in response to the device state images modified by the prospective network element failovers given the device configurations. It is then detected whether an acceptable performance is maintained in view of the prospective network element load and performance levels after failover. Thus, a predictive analysis can be performed in order to determine whether an SLA would be violated in view of potential failovers or errors with a given network configuration. When it is found that an SLA violation may occur, failover settings may be re-shuffled (i.e., reset) to a new configuration that avoids the potential SLA violations.
In the case of network maintenance, a re-convergence action may be pro-actively initiated to prevent outages. Effectively, the process is the same as above except that when the simulation determines that the failover plan is acceptable, it may format and send a simulated device failure message to a network manager in order to stimulate the re-convergence action before the users experience an outage. Since no element failure has occurred yet, paths that haven't been re-routed yet will not experience an outage—even if the convergence action takes minutes or hours. Once all of the re-routing or re-convergence is completed, the maintenance can be performed without interruption to the applications.
When a router 11-21 fails or is removed from the network for maintenance, the network is reconfigured so that the remaining devices take its place. The network devices may be grouped so that data paths using a direct link between routers 19 and 21 are re-routed to links (resources) between 19 and 12, and 12 and 21, for example. If two or more routers within the same group attempt to switch over to the same failover resources during the same time period, then the switchover may fail for one or all devices or congestion may result at the devices providing the failover resources thereby significantly degrading network performance. Due to the complexity of the network interconnections, however, it has not been possible to predict the impact on network performance of removing a device from service to perform maintenance.
Element managers 30 and 31 are coupled to a network manager application 33 for consolidating and analyzing data from the element managers and for determining any configuration changes that should be made in the network. Network manager 33 may be comprised of the Conscious™ traffic engineering system available from Zvolve Systems, Inc., for example, or other network level tools.
The present invention employs a simulator 34 coupled to network manager 33 and to a user interface 35. Simulator 34 generates a network model for predicting network performance based on actual and prospective (i.e., hypothetical or potential) states of the network elements. User interface 35 allows a network operator or other user to review the network model, to obtain predictions based on prospective actions and modifications of the network elements, and to implement modifications in the network element configurations (via known Network Manager functionality).
Improved network monitoring of the present invention employs a device state image 36 as shown in
Simulator 34 is shown in greater detail in
A preferred method of the present invention is shown in
In step 42, the simulator creates a network model and preferably simulates the current network performance as a baseline for comparing the effects of prospective changes to the device state images. Performance may be estimated as an average throughput or an average response time, for example. It may be desirable in the present invention to compare the estimated values with performance metrics required by a service level agreement (SLA) in order to detect potential nonconformance with the SLA, especially in deciding whether the impact of a prospective change is acceptable.
In step 43, one or more network elements are chosen for which maintenance actions are desired. Prospective failovers of the chosen devices are transmitted from the user interface to the simulator. Network performance is re-simulated in step 44 using the modified device state images for the chosen network elements together with changes in the network links resulting from a switchover to the failover devices. In other words, the device status of the chosen network element to be serviced is changed to offline and the link status of any affected devices are updated within the re-simulation.
A check is made in step 45 to determine whether network performance is acceptable. This may be determined by comparing the estimated performance with predetermined fixed performance levels. Alternatively, the difference between the baseline estimates and the new estimates can be compared to a threshold, with a difference below the threshold being acceptable and a difference above the threshold being an unacceptable change in performance. If performance is acceptable, then the chosen network element(s) are taken down for maintenance in step 46, thereby causing network traffic to switch to the failover devices. Maintenance actions are completed in step 47, and then the network elements are restored to a desired configuration (e.g., back to their original configuration or to another configuration).
If performance is found to be unacceptable in step 45, then an operator may choose a modified set of failover settings for one or more network elements in step 50. These potentially modified settings are evaluated by re-running the simulation in step 51 in an attempt to find an alternate network configuration in which acceptable network performance can be maintained while the chosen devices are taken down for maintenance. The potential modifications may be input by an operator via the user interface based on the operator's knowledge of the network and the results of simulations already performed. A check is made in step 52 to determine whether the maintenance action(s) could be performed on the hypothetically modified network while still maintaining acceptable performance. If not, then a return is made to step 50 to try different modifications to the failover settings. If step 52 determines that the modified failover settings do produce acceptable performance, then the new failover settings are transmitted to the affected network elements in step 53 via the network manager and the chosen network element is taken down for maintenance in step 46.
As demonstrated by the foregoing description, an inventive solution has been provided wherein data about network elements including their failover settings is collected at the element manager level, device state images are constructed and transmitted to a simulator, a potential resource change is transmitted to the simulator, a change in the network state or performance is determined by the simulator, and any necessary responsive action to the change is determined before the change occurs. The simulation allows a network operator to know in advance the result on network performance of initiating a specific maintenance action in the current network state. Further interactions or changes in network performance resulting from simultaneous fault conditions in network elements and maintenance actions in other network elements are made visible to the network operator, allowing countermeasures or mitigating actions to be taken.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US6108300||Aug 25, 1997||Aug 22, 2000||Cisco Technology, Inc||Method and apparatus for transparently providing a failover network device|
|US6295558||Aug 21, 1998||Sep 25, 2001||Hewlett-Packard Company||Automatic status polling failover or devices in a distributed network management hierarchy|
|US6477483 *||Jan 17, 2000||Nov 5, 2002||Mercury Interactive Corporation||Service for load testing a transactional server over the internet|
|US6519228 *||Dec 22, 1999||Feb 11, 2003||International Business Machines Corp.||System and method of operation for verifying and validating public switch telephone networks (PSTN) to (IP) network services|
|US6832184 *||Mar 2, 2000||Dec 14, 2004||International Business Machines Corporation||Intelligent work station simulation—generalized LAN frame generation simulation structure|
|US6874099 *||May 31, 2001||Mar 29, 2005||Sprint Communications Company L.P.||Method and software for testing and performance monitoring|
|US20010038471 *||Jan 5, 2001||Nov 8, 2001||Niraj Agrawal||Fault communication for network distributed restoration|
|US20020097672||Nov 5, 2001||Jul 25, 2002||Crescent Networks, Inc.||Redundant control architecture for a network device|
|US20030156549 *||Jan 9, 2003||Aug 21, 2003||Robert Binder||Method and system for evaluating wireless applications|
|US20040032857 *||Mar 19, 2003||Feb 19, 2004||Lcc International, Inc.||Methods, apparatus, and systems for simulation of mixed traffic in a wireless network|
|EP1073244A1 *||Jul 29, 1999||Jan 31, 2001||International Business Machines Corporation||Method and system for monitoring dynamic host configuration protocol (DHCP) service in an internet protocol network|
|WO2003039070A2 *||Nov 1, 2002||May 8, 2003||British Telecomm||Method and apparatus for analysing network robustness|
|1||*||Sai Kiran Reddy, G et al., "Reconfiguration based Failure Restoration in Wavelength-Routed WDM networks," Proceedings International Conference on Dependable Systems and Networks, 2000. Jun. 25, 2000. pp. 543-552.|
|2||Thomas Heim, Agilent Technologies, Meeting the Challenges of Integrated Networks, Nov. 12, 2001, pp. 1-35.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8363543 *||Nov 24, 2009||Jan 29, 2013||At&T Intellectual Property I, L.P.||Management of network capacity to mitigate degradation of network services during maintenance|
|US8509098 *||Apr 28, 2006||Aug 13, 2013||Alcatel Lucent||Method and apparatus for identifying network connectivity changes in dynamic networks|
|US8743742 *||Apr 16, 2012||Jun 3, 2014||Riverbed Technology, Inc.||System and method for modeling a system that comprises networks connected across a third party external network based on incomplete configuration data|
|US8775597 *||May 21, 2012||Jul 8, 2014||Eci Telecom Ltd.||Technique for management of communication networks|
|US8804490 *||Sep 19, 2011||Aug 12, 2014||Telefonaktiebolaget L M Ericsson (Publ)||Controller placement for fast failover in the split architecture|
|US8804500 *||Nov 19, 2012||Aug 12, 2014||At&T Intellectual Property I, L.P.||Management of network capacity to mitigate degradation of network services during maintenance|
|US8811212||Feb 22, 2012||Aug 19, 2014||Telefonaktiebolaget L M Ericsson (Publ)||Controller placement for fast failover in the split architecture|
|US8953441 *||Mar 15, 2013||Feb 10, 2015||Juniper Networks, Inc.||Re-routing network traffic after link failure|
|US8995249||Feb 13, 2013||Mar 31, 2015||Amazon Technologies, Inc.||Predicting route utilization and non-redundant failures in network environments|
|US9059928||Jul 9, 2014||Jun 16, 2015||Telefonaktiebolaget L M Ericsson (Publ)||Controller placement for fast failover in the split architecture|
|US9069611 *||Oct 10, 2011||Jun 30, 2015||Adaptive Computing Enterprises, Inc.||System and method for providing dynamic provisioning within a compute environment|
|US9137143||Nov 7, 2011||Sep 15, 2015||Intel Corporation||Progressive adaptive routing in a dragonfly processor interconnect network|
|US9185027||Nov 11, 2011||Nov 10, 2015||Telefonaktiebolaget L M Ericsson (Publ)||Method and apparatus for resilient routing of control traffic in a split-architecture system|
|US9210050 *||Jan 4, 2010||Dec 8, 2015||Centurylink Intellectual Property Llc||System and method for a testing vector and associated performance map|
|US9225591||Mar 3, 2015||Dec 29, 2015||Telefonaktiebolaget L M Ericsson (Publ)||Controller placement for fast failover in the split architecture|
|US20070253341 *||Apr 28, 2006||Nov 1, 2007||Lucent Technologies Inc.||Method and apparatus for identifying network connectivity changes in dynamic networks|
|US20110010585 *||Jan 13, 2011||Embarg Holdings Company, Llc||System and method for a testing vector and associated performance map|
|US20110122768 *||May 26, 2011||At&T Intellectual Property I, L.P.||Management of Network Capacity to Mitigate Degradation of Network Services During Maintenance|
|US20120005331 *||Jan 5, 2012||At&T Intellectual Property I, L.P.||Method and system to identify a source of signal impairment|
|US20120030685 *||Feb 2, 2012||Adaptive Computing Enterprises, Inc.||System and method for providing dynamic provisioning within a compute environment|
|US20120144065 *||Jun 7, 2012||Cray Inc.||Table-driven routing in a dragonfly processor interconnect network|
|US20120201168 *||Aug 9, 2012||Opnet Technologies, Inc.||System and method for modeling a system that comprises networks connected across a third party external network based on incomplete configuration data|
|US20120303780 *||May 21, 2012||Nov 29, 2012||Dukhovny Vadim||Technique for management of communication networks|
|US20130028073 *||Sep 19, 2011||Jan 31, 2013||Telefonaktiebolaget L M Ericsson (Publ)||Controller Placement for Fast Failover in the Split Architecture|
|US20130077480 *||Mar 28, 2013||At&T Intellectual Property I, L.P.||Management of network capacity to mitigate degradation of network services during maintenance|
|US20130329548 *||Mar 15, 2013||Dec 12, 2013||Harshad Bhaskar Nakil||Re-routing network traffic after link failure|
|U.S. Classification||370/244, 370/242, 379/29.02, 370/218|
|Mar 12, 2003||AS||Assignment|
Owner name: SPRINT COMMUNICATIONS COMPANY, LP, KANSAS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:COOK, FRED S.;REEL/FRAME:013867/0515
Effective date: 20030310
|Apr 24, 2015||REMI||Maintenance fee reminder mailed|
|Sep 13, 2015||LAPS||Lapse for failure to pay maintenance fees|
|Nov 3, 2015||FP||Expired due to failure to pay maintenance fee|
Effective date: 20150913